Video-based gait analysis using pose estimation can quantify gait differences among non-frail, pre-frail, and frail older adults

The study examined the use of video recording and human pose estimation technology to analyze gait in older adults. The research included 68 seniors divided into three groups: 25 non-frail, 25 pre-frail, and 18 frail individuals. Participants were recorded while walking at their self-selected pace and at a faster pace. Results showed that step time measurements using pose estimation were highly correlated with manual measurements (R² = 0.93 at self-selected pace and R² = 0.80 at fast pace). The method demonstrated the ability to identify differences in gait parameters among groups according to frailty status. This video-based gait analysis has potential for scalable gait tracking and frailty screening in older adults.